Abstract
Estimation of spectrum parameters of a stationary and homogeneous random field is examined in terms of a pair of fixed probes. The effect of finite probe volumes on estimators for determining the mean wave-number and the wave-number spectral width is studied by integrating the random field over the probe volumes. So long as this effect is unsignificant, the estimate biases tend to decrease with the probe separation, while the statistical errors are enhanced. Thus there exists an optimum probe separation. Various types of correlators may be used for evaluating the complex covariance of the probe signals easily. However a method of evaluation, which is analogous to the pulse-pair processing in radar systems, is preferred by reason of relatively high statistical accuracy.
